I've recently switched from FirefoxDriver to GhostDriver.
My test suite (roughlty around 150 end to end tests) used to take 25 minutes to run. Now with Ghostdriver it's taking 23 minutes. So in total I've gained 2 minutes of run time.
It's an improvement but not as huge as I would've expected. Is such a low gain normal ? Or should I expect a much higher cut down time from switching to an headless test runner ?
I'm using the .NET version of webdriver/ghostdriver.
I'd be really interested to compare this "benchmark" with anyone having recently switched to GhostDriver
GhostDriver won't offer any performance gain over ChromeDriver as the only thing it is not doing is rendering the graphics on the screen which Chrome does using a graphics driver and not the CPU.
The only benefit I see is not popping up a browser window and maybe when running on a CI server.
Some stats on my test run are on my blog: http://watirmelon.com/2013/02/05/watir-webdriver-with-ghostdriver-on-osx-headless-browser-testing/
I feel like this is bit vague question. For example, if your test contains lots of Thread.sleep() calls, then there will be no improvement to the tests.
My tests use ChromeDriver and I didn’t think about switching. But I feel like that simple switch will not gain you much, maybe also consider refactoring your code
来源：https://stackoverflow.com/questions/14194371/ghostdriver-actual-performance-gain